Mailing-List: contact reviews-help@spark.apache.org; run by ezmlm
Precedence: bulk
From: MechCoder <git@git.apache.org>
To: reviews@spark.apache.org
Reply-To: reviews@spark.apache.org
References: <git-pr-7854-spark@git.apache.org>
In-Reply-To: <git-pr-7854-spark@git.apache.org>
Subject: [GitHub] spark pull request: [SPARK-9525] [PySpark] [MLlib] Optimize
 Sparse...
Content-Type: text/plain
Message-Id: <20150911013048.8FAF7E1082@git1-us-west.apache.org>
Date: Fri, 11 Sep 2015 01:30:48 +0000 (UTC)

Github user MechCoder commented on a diff in the pull request:

    https://github.com/apache/spark/pull/7854#discussion_r39233952
  
    --- Diff: python/pyspark/mllib/linalg/__init__.py ---
    @@ -461,32 +461,41 @@ def __init__(self, size, *args):
             self.size = int(size)
             """ Size of the vector. """
             assert 1 <= len(args) <= 2, "must pass either 2 or 3 arguments"
    -        if len(args) == 1:
    -            pairs = args[0]
    -            if type(pairs) == dict:
    -                pairs = pairs.items()
    -            pairs = sorted(pairs)
    -            self.indices = np.array([p[0] for p in pairs], dtype=np.int32)
    -            """ A list of indices corresponding to active entries. """
    -            self.values = np.array([p[1] for p in pairs], dtype=np.float64)
    -            """ A list of values corresponding to active entries. """
    +        if isinstance(args[0], bytes):
    +            assert isinstance(args[1], bytes), "values should be string too"
    +            if args[0]:
    +                self.indices = np.frombuffer(args[0], np.int32)
    +                self.values = np.frombuffer(args[1], np.float64)
    +            else:
    +                # np.frombuffer() doesn't work well with empty string in older version
    +                self.indices = np.array([], dtype=np.int32)
    +                self.values = np.array([], dtype=np.float64)
             else:
    -            if isinstance(args[0], bytes):
    -                assert isinstance(args[1], bytes), "values should be string too"
    -                if args[0]:
    -                    self.indices = np.frombuffer(args[0], np.int32)
    -                    self.values = np.frombuffer(args[1], np.float64)
    -                else:
    -                    # np.frombuffer() doesn't work well with empty string in older version
    -                    self.indices = np.array([], dtype=np.int32)
    -                    self.values = np.array([], dtype=np.float64)
    +            if len(args) == 1:
    +                args = args[0]
    +                if isinstance(args, dict):
    +                    args = args.items()
    +                args = list(zip(*args))
    +
    +            # Handle empty args case.
    +            if len(args) == 0:
    +                indices = []
    +                values = []
                 else:
    -                self.indices = np.array(args[0], dtype=np.int32)
    -                self.values = np.array(args[1], dtype=np.float64)
    -            assert len(self.indices) == len(self.values), "index and value arrays not same length"
    -            for i in xrange(len(self.indices) - 1):
    -                if self.indices[i] >= self.indices[i + 1]:
    -                    raise TypeError("indices array must be sorted")
    --- End diff --
    
    hey, I'll update it definitely in a while (I sent you a mail regarding how much time I would be able to allocate from now on)


---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---

---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org